可以将文章内容翻译成中文,广告屏蔽插件可能会导致该功能失效(如失效，请关闭广告屏蔽插件后再试):
问题:
I have a label matrix with dimension (100*100), stored as a numpy array, and I would like to display the matrix with pyglet.
My original idea is to use this matrix to form a new pyglet image using function pyglet.image.ImageData(). It requres a buffer of the imagedata as an input, however I have no idea how to get a right formated buffer from the numpy array.
Any one have any idea?
ps. my current solution:
3d_label = numpy.empty([100,100,3]) 3d_label[:,:,0] = label * 255 # value range of label is [0,1] 3d_label[:,:,1] = label * 255 3d_label[:,:,2] = label * 255 image_data = ctypes.string_at(id(3d_label.tostring())+20, 100*100*3) image = pyglet.image.ImageData(100, 100, 'RGB', image_data, -100*3)
Any better way to construct a [100*100*3] matrix from 3 [100*100] matrix with numpy?
回答1:
I think what you are looking for is np.dstack (or more generally, np.concatenate):
label255=label*255 label3=numpy.dstack((label255,label255,label255))
This shows dstack produces the same array (label3) as your construction for label_3d:
import numpy as np label=np.random.random((100,100)) label255=label*255 label3=np.dstack((label255,label255,label255)) label_3d = np.empty([100,100,3]) label_3d[:,:,0] = label * 255 # value range of label is [0,1] label_3d[:,:,1] = label * 255 label_3d[:,:,2] = label * 255 print(np.all(label3==label_3d)) # True
PS. I'm not sure, but have you tried using label3.data instead of ctypes.string_at(id(label3.tostring())+20, 100*100*3) ?
回答2:
You can get the memory representation of your array with 3d_label.tostring().
The tostring() method allows you to change the memory ordering of the elements:
Parameters ---------- order : {'C', 'F', None}, optional Order of the data for multidimensional arrays: C, Fortran, or the same as for the original array.
PS: The 3d_label.data of ~unutbu requires less memory, since no string is constructed. However, it does not allow you to change the order in which the elements are output.
翻译